The short version. Most owners believe a big negative-keyword list protects them from wasted spend. Across 31 audited accounts, the size of that list has almost no relationship to how much budget gets wasted: the correlation is near zero (r = -0.10). Wasted spend is real, it is large, and negatives are not the lever that removes it.
The number you are hunting is not printed anywhere in Google Ads. It is found by analysis, not by opening a report.
There is no single figure, and anyone who quotes you one flat number is quoting a methodology, not your account. What the data does show is a range, and the range is wide. Measuring zero-conversion search-term spend across accounts in the same industry, the waste share stretches from roughly 6% in the cleanest account to nearly 40% in the messiest. In one set of six direct competitors, it ran 6.4% to 31%. In another niche, 13% to 40%.
Two things follow. First, waste has a floor. Even the best-run accounts carry a real slice of spend that never converts, so the goal is not zero, it is control. Second, the spread between the best and worst account is far larger than most owners assume, which means the money is findable. The difference between a 10% account and a 35% account is not luck. It is method.
This is the part that surprises people. If negatives were the fix, then accounts with more negatives would waste less. They do not. Across 31 accounts, the size of the negative list, measured against the account’s search-term volume, shows essentially no correlation with waste. The number is r = -0.10, which in plain terms means the relationship is close to random. An account with tens of thousands of negatives can sit right next to an account with almost none, at the same waste level.
The reason is mechanical, and it is the most common mistake I find. Most people add negatives the easy way: they open the search-terms report, tick a box on a query they do not like, and click add as negative. That adds the entire query as one negative. It blocks that exact string and nothing else. Change one word in the query and the negative no longer applies. So the list grows into the thousands while the actual leak keeps flowing, because the leak was never a single query. It was a word.
Here is the trap. Open your search-terms report, sort by cost, and the worst individual queries look almost reasonable. No single line screams waste. That is exactly why the money survives. The damage is not in any one query. It is in a word that appears across a hundred thousand different queries, each one too small to notice, adding up to a fortune.
In the corpus, generic informational words like free, how, code, pdf, generator, and online each bled between $0.4M and $2.6M in zero-conversion spend, and they showed up in 8 to 10 of every 10 verticals. You cannot see that by scrolling. You see it by running an n-gram analysis, which counts spend at the level of individual words instead of whole queries. One word, one filter, and a cluster of tens of thousands of queries resolves into a single number.
This is a correlation-versus-causation problem, the one every dashboard sets you up to fail. The size of your negative list and your waste level are two numbers that feel connected, so you assume one controls the other. The data says they barely move together. Piling on more negatives is not the cause of less waste. Finding the words that actually bleed is.
Profit Forensics
I examine one week of your Google Ads account and find the money it is losing, or prove it is clean. Signed, either way. For accounts spending $50,000 or more per month.
| Approach | What most owners do | What actually moves waste |
|---|---|---|
| Negatives | Add whole queries from the search-terms report | Cut the n-gram word behind thousands of queries |
| Diagnosis | Scroll and sort by cost | Count spend at the word level, verify conversions first |
| Relevance | Assume the query is the problem | Check query to ad to landing page for intent mismatch |
Researcher’s take
Every owner who calls me is looking for a secret. A hidden report in Google Ads, a setting nobody else knows, a screen with one big number that says wasted. It does not exist. Wasted spend is not a place you visit. It is a process. You verify the conversions are even real, you sort the search terms, you run an n-gram, and you watch one ordinary looking word turn out to be eating a hundred thousand queries. The account does not hide the money in a menu. It hides it in plain sight, spread too thin to see without the math.
Igor Ivitskiy, PhD, Doctor Ads
How much of my Google Ads budget is wasted?
There is no single number. Measured as zero-conversion search-term spend, waste runs from about 6% in a well-run account to nearly 40% in a poorly-run one within the same industry. The right question is not the average, it is where your account sits in that range and why.
Do negative keywords reduce wasted spend?
Far less than people think. Across 31 accounts, the size of the negative list shows almost no correlation with waste (r = -0.10). Negatives added as whole queries block one exact string and miss the word that leaks across thousands of queries. The fix is n-gram-level filtering, not list size.
Why can’t I see the wasted spend in my account?
Because it is not concentrated in any one query. It is spread across a hundred thousand small queries that share one ordinary word, so no single line looks bad. You surface it by counting spend at the word level with an n-gram analysis, not by scrolling the search-terms report.
What is an n-gram analysis in Google Ads?
It counts spend and conversions by individual words or short word-groups across all of your search terms, instead of by whole queries. That reveals a single word quietly draining budget across thousands of queries, which is invisible when you look query by query.
Figures come from a forensic analysis of 31 Google Ads accounts representing $133M in managed spend, September 2024 to February 2025. Waste is defined as spend on search terms with zero recorded conversions. The negatives-to-waste correlation is measured as negative-list size relative to search-term volume against waste share across the 31 accounts. Word-level bleed figures are zero-conversion spend attributed to individual generic terms across verticals. Related reading: bid adjustments and Target CPA bidding.